How to reduce huge dimensions
Hello, I'm having a lot of issues with workspace size, with models going above 150/200gb in size and I'm not sure how to reduce it, to give some context, I have three main lists (size considering subtotals):
- Products (~1,5k)
- Countries (>100)
- Customers (>270)
The issue is that if we apply all of them to a module what we get is an average of 563million cells per line item, the huge problem that I recognize that not all products are sold in all countries and not to all customers, but when we apply them to the models we get all possibilities, even if they don't make sense…
Is it possible to have only customers from an specific country appearing in the correct country and products that are only sold in that country appearing there? It would help a lot on the model size.
The only issue is that we can't merge the hierarchy, i.e., assign the product under the country on the countries hierarchy, we need the three of them to be separate lists.
Is it possible?
Best Answer
-
You can have the 3 lists as they are and use them elsewhere, but you can absolutely have a hierarchical/composite list where you have:
Countries
Customers
ProductsAnd, I imagine this could be true in that you sell a lot of the same products to different customers, have the hierarchy stop at Customers (Countries and Customers are the hierarchy) and then have products by itself.
1
Answers
-
Take a look at this resource
You could keep it as three separate lists, but really you need to use the Polaris sparse engine for this.
You will find much more benefit from creating the correct hierarchy based on the attributes of each dimension.0 -
Thanks for the idea @rob_marshall. I was hesitant to merge them, but indeed, it seems to be the best option.
0